A control plane for distributed AI could help applications decide which tasks to run on a device, at the edge or in the cloud.
Seismora Inc. is developing networking technology to coordinate AI workloads across different providers and computing environments, according to Vito Palermo (pictured), founder and chief executive officer of Seismora. The aim is to let developers use resources across those environments without building the routing logic into each application.
“The AI boom creates a scenario where we’re moving from human-to-system communication to machine-to-machine,” he said. “What we’re building at Seismora is an intelligent control plane that allows AI traffic to move across the network, regardless of the network, regardless of who the neocloud is or the hyperscaler, or even the devices themselves.”
Palermo spoke with John Furrier for theCUBE + NYSE Wired: AI Luminaries interview series on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed the infrastructure needed to support AI applications across distributed networks. (* Disclosure below.)
Control plane for distributed systems
Different parts of an AI application may require different computing resources, according to Palermo. Seismora’s control plane coordinates that work across devices, edge infrastructure and cloud providers.
“In a distributed control plane, the system becomes smart enough to say, ‘Okay, based on the intent of this user for this application, we can do X amount of work here on the iPhone, we can do Y amount of work here at the edge,’” Palermo told theCUBE. “For specific parts of that, maybe you want to build a digital twin using Nvidia [Corp.’s] Omniverse. Let’s go do that in a neocloud. And the control plane automatically makes those decisions and optimizes the cost, too.”
Palermo pointed to Stripe Inc.’s agreement to acquire OpenRouter Inc., an AI model routing platform, as evidence of growing interest in infrastructure for machine-to-machine transactions. The August announcement involved a price reportedly around $7.5 billion.
“You want to be heterogeneous because, frankly, the enterprise today, I’m sure they have multiple neoclouds they work with, they have multiple models they work with,” Palermo said. “My problem is routing across all of these providers, and that’s the focus.”
Seismora calls its approach “cognitive routing”: choosing where AI work runs based on capability, cost, latency and policy constraints. Its intended customers are developers building agentic applications for end users, rather than consumers themselves, according to Palermo.
“We view a world where, for example, [there are] open-weight models from providers like Fireworks [Fireworks.ai Inc.], contextual knowledge from Neo4j,” Palermo explained. “Then, having compute and storage distributed across the edge, across the enterprise, and to be able to move intelligence where it makes sense.”
Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of theCUBE + NYSE Wired: AI Luminaries interview series:
(* Disclosure: TheCUBE is a paid media partner for theCUBE + NYSE Wired: AI Luminaries interview series. Neither Neo4j, the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)